hammsandwich

Anyone have thoughts on this? Really curious to figure this out so I do not have to shut down this conatiner whenever I need to start up other VMs.

Hello, This container is somehow utilizing huge pages on my Unraid server. I have enabled hugepages through the normal means of: append hugepagesz=2M hugepages=8194 This allocates 16GB of RAM to huge pages. Then have various VMs that leverage these hugepages via the following XML. <memoryBacking> <hugepages/> <nosharepages/> </memoryBacking> I noticed a few months back that at certain times I was unable to start 16GB worth of VMs. I began shutting down containers, which would then allow me to fully populate all 16GB of hugepage RAM, so I then began the process of narrowing down which container was causing the issue. It appears to be this Nextcloud container. I used this command to see what was grabbing hugepages: grep huge /proc/*/numa_maps When Nextcloudis running, this command outputs: grep huge /proc/*/numa_maps /proc/24581/numa_maps:145fd0a00000 default file=/anon_hugepage\040(deleted) huge dirty=3 mapmax=3 N0=3 kernelpagesize_kB=2048 /proc/24606/numa_maps:145fd0a00000 default file=/anon_hugepage\040(deleted) huge dirty=11 mapmax=3 N0=11 kernelpagesize_kB=2048 /proc/24607/numa_maps:145fd0a00000 default file=/anon_hugepage\040(deleted) huge dirty=11 mapmax=3 N0=11 kernelpagesize_kB=2048 Upon shutting down this container, there is no output to the command (if all my VMs are also off). I have started every single one of my containers one by one, and this is the only one that results in this behavior. Is this intentional? If so, I feel that it should be documented as many people use hugepages for VFIO passthrough performance. If not intended, is the proper place to for support for this issue? Thanks in advance.

Confirming based on your github documentation the Zen 3 platform, i.e. 5800X is currently not supported?

Sure thing, Just to confirm, I don't haveto manually start the daemon, just run the corefreq-cli?

Set BIOS OC back to AUTO, only have RAM OC applied. I am unable to change my CPU Freq past 3800 Mhz on my 5800X. When I attempt to load the CPU-FREQ driver or the CPU-IDLE driver, I get this following message: Here is what my Max Freq option looks like The rest of the freq options looks similar, they all show the same freq even thought i have disabled turbo through the plugin. Uploading diagnostics... hammsandwich-diagnostics-20210606-0843.zip

It is not multiple accounts. but multiple share types. For example, share "ABC" is set to Public, and share "XYZ" is set to Private, with R/W perms for user "123". If I browse via Windows to \\unraid\ABC, I am immediately allowed in as an anonymous user and can read and write to share "ABC" perfectly fine. Then I want to authenticate to share "XYZ" to work on more sensitive data. I get prompted with username/password, and when entered correctly, Windows throws an error saying it cannot access different shares with multiple accounts on the same file server. I am assuming this means it is counting the Anonymous user as a user. I have been messing around this this all day and currently cannot reproduce this issue but am unable to at the moment. It is possible that changing the permissions of each share from private to public and back to private has resolved my issues... This has been going on for the past 2 years or so I have finally gotten tired of it and am trying to fix the issue.

I have noticed I set shares to public they work without issue. When a share is private, I cannot copy anything larger than a few MB without a transfer fail. Retry works about 1/5 times. This is prevalent on multiple Win10 machines on my network, all various versions with various hardware types. Unfortunately having a mix of Public and Private shares also does not work because Windows does not allow for multiple connections to the same SMB servers with different accounts simultaneously.

You are absolutely correct. After enabling additional firewall logging and reinstalling the plugin and rebooting, I am seeing Blocks on the firewall from Unraid to 8.8.8.8 on ICMP. Added to outbound rules and that helped with speed of install and booting. I plan on disabling BIOS OC this weekend to see if that resolves my VM stability issue.

Oh I apologize I didn't understand the question initially. No, I run both on other hardware with no reliance on Unraid, so rebooting Unraid would have no impact on internet access to any devices on my network.

I don't see any blocks at the firewall or PiHole.. I allow all github traffic and use stock PiHole so I do not think these are causing issues but will check again.

I do have a firewall between Unraid and the internet, just check the logs and I allow github traffic. I see no denies. I also have PiHole, and just mainly the default block lists with a few regex filters, but also checked those logs and there are no blocks either.

I have not used the plugin yet to set OC or anything, I am still doing an all-core OC/undervolt in the BIOS, which has been stable for a month+ after very thorough testing. My installing the plugin, my daily driver Win10 VM with various passthroughs just randomly...does something odd. All my monitors turn off, but Unraid reports its still up. Restarting the VM via Unraid does not work; I have to force stop and power it on again. I have 2C/4T for Unraid, then the remaining 6C/12T isolated from Unraid for VMs. I noticed just while poking around Idle drivers and other various drivers were not loaded according to corefreq. Attaching diag before uninstalling again to ensure this is related to corefreq. hammsandwich-diagnostics-20210602-0956.zip

It is installed and I can see the dashboard. I did reboot my server afterward, and it took noticeably longer to reboot than normal. I uninstalled the plugin and rebooted, and it went quicker. Finally I reinstalled the plugin (took a while again) and rebooted, and the boot from GRUB takes about twice as long until I can see the dashboard.

Thanks for putting hard work into making this plugin. EDIT: DISREGARD it took way longer than I expected to install. When I attempt to install it, the install never complete. I see the following: plugin: installing: https://raw.githubusercontent.com/ich777/unraid-corefreq/master/corefreq.plg plugin: downloading https://raw.githubusercontent.com/ich777/unraid-corefreq/master/corefreq.plg plugin: downloading: https://raw.githubusercontent.com/ich777/unraid-corefreq/master/corefreq.plg ... done I have reattempted the install multiple times, but all attempts result with the same. Occasionally there will be an update button next to the install button, but that takes me to the plugins tab and I have confirmed this plugin is not there. I have a Ryzen 5800X in a ASRock x470 Taichi.

Thanks! I can't wait to give this a shot!

Requesting adding support for corefreq and modules. https://github.com/cyring/CoreFreq

It finally connected and updated. Very odd. Might need to check my DNS infra. Thanks.

Starting my Valheim container this morning hoping to pull down the update, the server never loads and I just get this repeating in my logs: Waiting for user info...'---Checking if UID: 99 matches user--- usermod: no changes ---Checking if GID: 100 matches user--- usermod: no changes ---Setting umask to 000--- ---Checking for optional scripts--- ---No optional script found, continuing--- ---Starting...--- ---Update SteamCMD--- Redirecting stderr to '/serverdata/serverfiles/Steam/logs/stderr.txt' Looks like steam didn't shutdown cleanly, scheduling immediate update check [ 0%] Checking for available updates... [----] Verifying installation... Steam Console Client (c) Valve Corporation -- type 'quit' to exit -- Loading Steam API...OK. Connecting anonymously to Steam Public...Retrying. . . Warning: failed to init SDL thread priority manager: SDL not found Retrying. . . Retrying. . . Retrying. . . Logged in OK Waiting for user info...OK ---Update Server--- Redirecting stderr to '/serverdata/serverfiles/Steam/logs/stderr.txt' [ 0%] Checking for available updates... [----] Verifying installation... Steam Console Client (c) Valve Corporation -- type 'quit' to exit -- Loading Steam API...OK. Connecting anonymously to Steam Public...Retrying. . . Retrying. . . Warning: failed to init SDL thread priority manager: SDL not found Retrying. . . Retrying. . . Logged in OK Session terminated, killing shell... ...killed. Waiting for user info...'---Checking if UID: 99 matches user--- usermod: no changes ---Checking if GID: 100 matches user--- usermod: no changes ---Setting umask to 000--- ---Checking for optional scripts--- ---No optional script found, continuing--- ---Starting...--- ---Update SteamCMD--- Redirecting stderr to '/serverdata/serverfiles/Steam/logs/stderr.txt' Looks like steam didn't shutdown cleanly, scheduling immediate update check [ 0%] Checking for available updates... [----] Verifying installation... Steam Console Client (c) Valve Corporation -- type 'quit' to exit -- Loading Steam API...OK.

Yes, the container has its own IP via macvlan, so its easy to distinguish this vs other Unraid traffic. Can you elaborate? I use a Palo Alto virtual firewall, and it does all my filtering and and traffic logging.

I have the Valheim server running on a specific IP, which I have NAT'ed and opened up externally on my firewall. It would be strange for SIP to be coming from your server, would it not? This is Valheim, I do not think there is voice chat enabled in the game. As soon as I start the container I see it trying to reach out to various IPs: 172.17.229.225 192.168.254.19 10.0.0.200 192.168.3.61 192.168.153.1 192.168.237.1 These are all internal IPs, and I do not have any of these network ranges on my internal network, so I see these in my firewall logs as drops. Could be internal IPs of people connecting to my server? I find this unlikely as I only have 4 external people connecting. Just some of the ports I am seeing, all UDP: 54678, 54676, 65468, 51777, 51779, 54056, 56492, 56496, 61559, 61557

Thanks for all the hard work on these dockers. I am host a private server and only allow my friends to connect via strict firewall rules (they give me their public IP to whitelist). I am seeing something pretty odd in my firewall logs. I am seeing lots of firewall from my server to internal IP addresses that are not on my network. 10.X addresses, 172.17 addresses. Since these addresses do not exist, all this traffic is getting dropped by the firewall. This traffic is all high UDP ports, 56000+. From local clients, I can see traffic passing between client and server on these, and wonder if this is normal. EDIT: I have the Valheim server running on a specific IP, which I have NAT'ed and opened up externally on my firewall.

Has anyone had luck with PCI passthrough and IOMMU groups on these new BIOS's? I head there were issues. I have a Win10 VM with multiple PCI passthrough devices, including NVMe, NVIDIA GPU, and onboard USB controller (with ACS patch).

As this docker relies on an older version of the upstream lancache dockers, and he is not tagging the builds or pulling tagged versions, I do not think this is currently possible.

Looks like a new version was built off a fork to remove snip proxy. Any pull this yet and see if it works? I am not in front of a computer to test.

Same here after latest update. Same error messages filling up nginx error logs.